winchester6788

I had a similar problem, and when claude code (or codex) is running in sandbox, i wanted to put a cap or get notified on large contexts.

especially, because once x0K words crossed, the output becomes worser.

https://github.com/quilrai/LLMWatcher

made this mac app for the same purpose. any thoughts would be appreciated

interesting that you chose to go the MITM way.

https://github.com/quilrai/LLMWatcher

here is my take on the same thing, but as a mac app and using BASE_URL for intercepting codex, claude code and hooks for cursor.

Only reason I use searchable logs is, when I roll back something in claude conversation but later want to refer it.

Apart from that, may be somebody might find it useful for observing their usage patterns etc.

No grouping or correlation right now. Wanted to get some feedback/ opinions first

Author of NudeNet here.

I just scraped data from reddit and other sources so i could build a nsfw classifier and chose to open source the data and the model for general good.

Note that i was a 1 year experienced engineer working solely on this project in my free time, so it was basically impossible for me to review or clear out the few csam images in the 100,000+ images in the dataset.

Although, now i wonder if i should never have open sourced the data. Would have avoided lot of these issues.

Ooh, love the UX, the heat meter and contrarian detection are very nice ideas.

I recently released https://trackernews.app/browse which surfaces out and groups content from hn, reddit .. for user defined topics and extracts user defined structured data along with summaries.

a tangential take on similar problem with focus on grouping posts by topics.

would love to get any feedback on the UX/ understandability of the interface

full of false positives though. but definitely good for some types of entities and regexes

This holds true in most other countries as well. Gore/ chopping of appendages is happily accepted and enjoyed (in movies, games etc) by all of India, whilst a simple kiss can be a taboo/ issue.

I made https://github.com/notAI-tech/fastDeploy for ease of use of deploying Deep learning models.

Except me and the team at my company nobody uses it.

After I made it lot of solutions came in this space, but IMO this is the easiest one to use still.

The model is trained by giving pairs of examples. So that it learns to generate an internal representation of images, texts or whatever inputs.

There are other ways to train, but this is the basic example.

2 chatgpt instances controlling the dual inputs would be funny

They both solve different issues. This is a library akin to huggingface transformers, while tensorflow js is akin to tensorflow or pytorch

Just wanted to drop in and appreciate your work.

It's documented very well, extremely easy to use and the pro version is priced very well.

> The 2021 M1 Air is a slightly chunkier fanless machine that runs circles around Intel laptops with desktop-class processors, at lower cost and up to triple/quadruple the battery life. It's unlike any Intel laptop you'll find on the market today.

Second this. I got the cheapest M1 macbook air in december and it really is streets ahead of competetion in terms of performace/ battery life.

The new macbook air is actually decent for gaming.

On the intel macbook airs (and on most pros without gpus) playing CS:GO at full resolution was a bad experience (overheating and <60fps).

I got the cheapest M1 macbook air and it plays CS:GO with no issues and without heating up (even when the ambient temperature in my city is ~38 Celsius)

If you can position the camera so that only the "counter" is visible from it, a much easier option is to use the pre-trained yolov4.

https://gist.github.com/YashasSamaga/e2b19a6807a13046e399f4b... (download links for yolov4.weights is at https://github.com/AlexeyAB/darknet)

Using this, you will be able to detect if a cat is present in your image.

Length of given text after encoding it with their BPE. In general, you can expect it to be 1.2x-2x the len(words) in your text.

> The crossover point where it's worth buying is where 1500 + 0.14h = 0.45h, about 5000 hours or 30 weeks of training.

1500 + 0.14h should be equal to 0.45h * 4. A 3080 will at the least be 2x faster for general training than one k80.

> In fact, most data science and ML engineers are quite skilled in systems engineering, because you have to do so much work with GPU hardware issues, underlying scientific package management, efficient data transportation, etc.

Unless your data scientists are expected to build the machines they use, they won't be dealing with any hardware issues at all.

Literally every data scientist at big companies use pre-configured vms/notebooks in cloud.

I strongly oppose this but it feels like we are heading this way.

I tried to collect as much data as possible for various body/ ethnicity types. The original data collection process is described in https://medium.com/@praneethbedapudi/nudenet-an-ensemble-of-...

Sorry for the delayed response. "COVERED_FEET" should have just been "covered Feet; Any gender". My description was bad/wrong there.